Cover Image
close this bookCauses and Consequences of Intrauterine Growth Retardation, Proceedings of an IDECG Workshop, November 1996, Baton Rouge, USA, Supplement of the European Journal of Clinical Nutrition (IDECG, 1996, 100 p.)
close this folderNutritional interventions to prevent intrauterine growth retardation: Evidence from randomized controlled trials
View the document(introduction...)
View the documentIntroduction
View the documentResults
View the documentDiscussion
View the documentConclusions
View the documentAnnex: Systematic reviews included
View the documentReferences
View the documentDiscussion

Results

1. Non-nutritional interventions

Based on our comprehensive review (Gglu et al, 1997), only 2 out of 24 non-nutritional interventions - smoking cessation and antimalarial chemoprophylaxis in primigravidae - are likely to be beneficial. Strategies to reduce smoking during pregnancy have been collectively analyzed in one systematic review, including 5 trials, for medical outcomes as well as separately for smoking rates (Lumley, 1995). In general, behavioral strategies have been found to be more effective in reducing smoking than advice and feedback. They are associated with increased birth weight and lower rate of term-LBW (typical odds ratio: 0.80; 95% CI: 0.65-0.98). No effect was detected on perinatal deaths. These effects have been greater in more compliant groups, and those groups who were more successful in stopping smoking showed the highest mean birth weight gains.

The systematic review on the use of antimalarial drug chemoprophylaxis in endemic areas includes 11 trials with data from more than 3,000 women on perinatal outcomes (Garner and Gglu, 1996). These studies had variable designs with different dosages and some using additional iron and folic acid supplements. Overall, malaria chemoprophylaxis was associated with higher maternal hemoglobin levels and birth weights. These effects were also more prominent in primigravidae, who are known to be more susceptible, showing an increase in mean birth weight of 112 g (95% CI: 41-183 g) in the treated group. However, the results for primigravidae alone are based only on 4 trials of small sample size (Garner and Gglu, 1996).

2. Nutritional interventions

There are 12 nutritional interventions during pregnancy aimed at preventing impaired fetal growth, including nutritional advice, nutritional supplementation (protein/energy, vitamin/minerals, and fish oil), and nutritional interventions for the prevention of hypertensive disorders (routine calcium and low versus high salt intake). Table 1 presents, for 11 interventions, the number of trials included in the systematic review, the type of intervention, characteristics of participants, and setting of the trials. Table 2 presents similar information for balanced protein/energy supplementation. Table 3 presents the results of the systematic reviews categorized as interventions likely to be beneficial, interventions of unknown effectiveness which merit further research, and interventions unlikely to be beneficial.

2.1. Nutritional interventions likely to be beneficial

Only one of the 12 nutritional interventions-balanced protein/energy supplementation in pregnancy-shows a reduction of borderline statistical significance in the incidence of SGA. The trials on balanced protein/energy supplementation (i.e., a supplement in which protein accounts for < 25% of the total energy content), generally performed in underprivileged populations, assessed the effects of providing pregnant women with nutritional supplements of energy (usually plus protein) on maternal weight gain, fetal growth, gestation duration and fetal/infant morbidity and mortality. Data on SGA are available from 7 of 13 trials included in this systematic review. These are those from Taiwan (Blackwell et al, 1973), Wales (Elwood et al, 1981), India (Girija et al, 1984), East Java (Kardjati et al, 1988), Bogota (More et al, 1978, 1979), Harlem (Rush et al, 1980a,b), and Thailand (Tontisirin et al, 1986) (Table 2). The overall methodological quality of the trials is variable, the major weakness being the failure to provide details of the randomization procedure and the exclusion after randomization of women who were unable or unwilling to take their allocated supplements. The degree of compliance and substitution of the normal diet may also have been quite variable across the trials (Kramer, 1997).

Supplementation was associated with increases in maternal weight gain and mean birth weight, and a decrease in the number of SGA babies of borderline statistical significance (Table 3 and Figure 1). However, there were certain inconsistencies, such as the improvement in fetal growth being smaller in the more under-nourished women, which is attributed to poor compliance or additional dietary supplements these women may have received. Another possibility is that under-nourished women are chronically adapted and do not respond to acute increases in intake. Statistically non-significant reductions in stillbirths and neonatal death are based on only two trials, but may be important if real (Kramer, 1997). A new community-based trial in rural Gambia, recently made available and thus not included in the calculations of the typical odds ratio, shows that supplementation significantly increased birthweight: which merit further research hungry season +201 g (p< 0.001), all year +136 g (p < 0.001). The odds ratio for LBW babies in supplemented women was 0.61 (95% CI: 0.47-0.79), supporting the results of the systematic review (Ceesay et al, 1997). Supplementation also significantly reduced perinatal mortality.

2.2. Nutritional interventions of unknown effectiveness which merit further research

A number of systematic reviews have examined vitamin/mineral supplementation (Table 1). The systematic review on routine zinc supplementation (between 20-25 mg per day) includes 4 trials (Goldenberg et al, 1995; Simmer et al, 1991; Mahomed et al, 1989; Hunt et al, 1984) but the effect on fetal growth is dominated by the results Of the large Alabama trial (Goldenberg et al, 1995). The overall methodological quality of the double blinded trials included in the review is high and the typical odds ratio shows a 23% statistically non-significant reduction in incidence of term-LBW (Table 3). There are, however, inconsistencies of results across trials. For example, a highly significant reduction in IUGR with supplementation (odds ratio 0.21; 95% CI: 0.05-0.96) was shown by the London trial (Simmer et al, 1991), while in Bristol (Mahomed et al, 1989) results failed to demonstrate any, significant difference in mean birthweights between the two groups. These conflicting results could be explained by the trials being performed in populations with varied zinc nutritional status and at variable risk for pregnancy out-comes. A most recent RCT, not included in the systematic review, conducted in a normal healthy population, did not demonstrate any effect on the incidence of SGA. Unfortunately, the methodology of this trial is questionable as the authors excluded from the analysis 794 women of the 2000 enrolled "because they forgot to take the tablets regularly" (Jn et al, 1996).

Table 1. Characteristics of randomized controlled trials included in systematic reviews of nutritional interventions reporting impaired fetal growth outcomes.

Systematic review

Number of trials

Type of intervention

Participants and setting

Comments

1. CARE AND ADVICE DURING PREGNANCY

Nutritional advice

1

Kafatos et al, 1989: Experimental: nutrition education to improve nutrient value of diet through home visits every 2 weeks by trained nurses; Control: no intervention

568 low socioeconomic status well-nourished pregnant women in the rural county of Florina, northern Greece, the majority enrolled < 21 wks and all before 27 wks gestation

Overall methodological quality not high; randomization by clinic but analysis based on individual subjects; dietary intake unblinded and energy intake higher in experimental group prior to intervention; inconsistencies in results and sample sizes for related outcomes

2. NUTRITION SUPPLEMENTATION

2.1. Protein/energy

Isocaloric balanced protein

1

Mardones et al, 1988: Experimental: supplement of powdered milk containing 498 kcal, 27.9% protein, 26% fat, 38% carbohydrates (per 100g of powder) plus vitamins/minerals; Control: milk based fortified product containing 470 kcal, 14.5% protein, 21% fat, 56% carbohydrates plus vitamins/minerals

1135 low income underweight pregnant women attending 9 prenatal clinics in the city of Santiago, Chile

Large losses to follow-up evenly distributed in the 2 groups; alternate treatment allocation but no evidence of bias; differences other than protein content between 2 trial supplements

High protein

1

Rush et al, 1980a,b: Experimental: 'supplement' of 470 kcal, 40g protein per day, plus vitamins/minerals; Control 'complement' of 322 kcal, 6g protein per day, plus vitamins/minerals

1051 low income black women in Harlem, New York City, ‘at risk’ for low birth weight based on established criteria

Overall high quality trial but almost no data reported on ~ 25% of randomized women who failed to comply, dropped out, or moved away

2.2 Vitamin and minerals

Vitamin D

1

Brooke et al, 1980: Experimental: Vitamin D 1000 IU/day throughout the last trimester until term Control: placebo

126 first generation immigrants pregnant Asian women living in London

Trial run over 2 years to avoid confounding due to variations in sunlight hours; preterm babies, congenital malformations, and maternal illnesses likely to affect fetal growth excluded but not clear if before or after randomization; low sample size

Routine zinc

4

Goldenberg et al, 1995: supplement of 25mg daily versus placebo;
Simmer et al, 1991: supplement of 22.5mg daily versus placebo;
Mahomed et al, 1989: supplement of 20mg daily versus placebo;
Hunt et al, 1984: supplement of 20 mg daily versus placebo

Goldenberg: 580 healthy African-American pregnant women with plasma zinc levels below the median at enrollment in prenatal care in Alabama, USA;
Simmer: 56 pregnant women ‘at risk' of delivering an SGA baby in London, UK;
Mahomed: 494 women < 20 weeks pregnant booking for delivery in Bristol, UK;
Hunt: 213 Hispanic low-income women attending a prenatal clinic in Los Angeles, USA

Overall quality of the 4 trials is high; all are double blind with few exclusions after entry; trials have been performed in populations with varied zinc nutritional status and risk for poor fetal growth which may in part explain conflicting results; in general, sample sizes inadequate to confirm or reject an effect on any specific pregnancy outcome

Routine magnesium

2

Kovacs et al, 1988: magnesium aspartate in a daily dose of 15 mmol (MG 5-Longoral) versus placebo;
Spatling & Spatling, 1988: magnesium-aspartate-hydrochloride in a daily dose of 15 mmol versus a placebo containing 13.5 mmol of aspartic acid per day

Kovacs: 985 well- nourished pregnant women in Germany;
Spatling: 568 low- and high- risk women attending out- patient clinic, Department of Obstetrics of University of Zurich, Switzerland

Trials have high numbers of exclusions and limitations in the randomization procedures (eg, the Spatling trial allocated the two groups according to the women's birthdates)

Routine iron

2

Dawson & McGanity, 1987: daily supplementation of one tablet multivitamin multimineral with 65 mg of iron versus one tablet multivitamin without iron;
Hemminki & Rimpel1991: for subjects in routine Fe group a single daily dose of 100 mg elemental Fe was suggested, but the choice of the drug and dosage were determined by the 51 midwives; the mean dose was 124 mg as calculated by information given by the mothers, subjects in selective group received 50 mg ferrous sulfate twice a day for 2 months if Hb < 100 g/L (criteria changed to 105 later in the study)

Dawson: 72 healthy young white and black women, < 16 wks pregnant, normal hematologic status and 'low-risk' singleton pregnancies at Univ. of Texas, USA;
Hemminki: multi-center trial involving 2944 healthy women £ 16 weeks gestation attending 27 maternity centers in Tampere, Finland

Dawson: Small sample size and large losses to follow-up; 43% of non-iron subjects were given 110 mg iron daily because Hb £ 11.0g/dL
Hemminki: Possible bias due to nonblinding of the study; iron dosage and drug not standardized for all study subjects; criteria for selective iron supplementation was changed in the middle of the study

Routine folate

5

Fleming et al, 1968: one tablet 5 mg folic acid administered forthrightly during the second and weekly during the third trimester of pregnancy versus one tablet of lactose base and colouring matter the same manner;
Baumslag et al, 1970: Group 1: 200 mg of iron; Group 2: 5 mg folic acid daily + iron; Group 3: 50µg vit B12 + folic acid + iron;
Giles et al, 1971: 5 mg folic acid daily + 200mg iron versus placebo + 200 mg iron;
Iyengar, 1971: Group 1: 100µg folic acid + 60 mg elemental iron daily; Group 2: 200 µg folic acid + 60 mg elemental iron daily; Group 3: 300µg folic acid + 60mg elemental iron daily; Control group: 60 mg elemental iron only;
Rolschau et al, 1979: Experimental: tablets containing ordinary vitamins, 250 mg ferrofumarate and 5 mg folic acid, Control: similar tablets but without the folic acid.

Fleming: 75 primigravida, < 26 wks gestation, with packed cell volume (PCV) 27% or more, attending in antenatal care at Ibadan, Nigeria;
Baumslag: 183 pregnant Bantu with diets low in folates and 172 White pregnant women with average Western diets in Johannesburg, South Africa;
Giles: 692 pregnant women attending antenatal clinic at Melbourne, Australia;
Iyengar: 200 low SES women between 20-24 weeks pregnancy attending nutrition outpatient clinic at Hospital in Hyderabad, India;
Rolschau: 40 Danish women with normal pregnancy at Odense University Hospital, Denmark

The quality of these trials is not good, populations poorly defined and limitations in randomization procedures; conducted in populations with routine iron supplementation; in some trials there are high numbers of exclusions after entry to the trial

2.3 Others

Prophylactic fish oil

1

Onwude et al, 1995: Treatment: 2.7g MaxEpa daily (1.62g of eicosapentaenoic acid and 1.08g of docosahexaenoic acid); Placebo: matching air-filled capsules

233 pregnant women at high of risk of developing proteinuric or non-proteinuric pregnancy induced hypertension or asymmetrical IUGR, attending antenatal clinic of St James's University Hospital, Leeds, UK


Fish oil versus placebo

1

Olsen et al, 1992: Women in week 30 of pregnancy were randomly assigned in a ratio of 2/1/1 to fish oil (four 1 g Pikasol capsules [containing 2.7g n-3 fatty acids] per day), olive oil (four 1 g capsules per day), or no supplement

533 healthy pregnant women attending the main midwife clinic in the city of Aarhus, Denmark

Well designed trial but there is some evidence that women allocated to olive oil were less likely to increase their fish consumption than women allocated to no oil; olive oil may also have effects so far not reported

Fish oil versus of olive oil evidence

1

Olsen et al, 1992: Women in week 30 pregnancy were randomly assigned in a ratio of 2/1/1 to fish oil (four 1 g Pikasol capsules [containing 2.7g n-3 fatty acids] per day), olive oil (four 1 g capsules per day), or no supplement

533 healthy pregnant women attending the main midwife clinic in the city of Aarhus, Denmark

Well designed trial but there is some that women allocated to olive oil were less likely to increase their fish consumption than women allocated to no oil; olive oil may also have effects so far not reported

3. PREVENTION HYPERTENSIVE DISORDERS

Routine calcium

5

Villar et al, 1987: Experimental: calcium 1.5 g day; Control: placebo;
Villar & Repke, 1990b: Experimental: calcium 2g day; Control: placebo;
Belizan et al, 1991: Experimental: calcium 2g/day; Control: placebo;
Cong et al, 1993: Experimental: Group A: calcium 1g/day, Group B: calcium 2g/day; Control: no calcium supplement;
Sanchez-Ramos et al, 1994: Experimental: calcium 2g/day; Control: placebo

Villar (1987): 52 healthy nulliparous or primiparous women (34 black in USA and 18 white in Argentina) 18-30 years, 26 weeks;
Villar (1990b): 190 healthy-pregnant adolescents (£ 17 years) in USA, 23 weeks;
Belizan (1991): 1194 healthy nulliparous women in Argentina, 20 weeks;
Cong (1993): 150 healthy pregnant women in China, 22 weeks;
Sanchez-Ramos (1994): 63 angiotensin-sensitive nulliparous women in USA, 20-24 weeks

Overall these trials are well-designed and conducted, most are double-blind and placebo-controlled; but usually have small sample sizes; women recruited were usually in their first pregnancy, but few had other risk factors for hypertensive disorders of pregnancy; effects derived from the individual trials are compatible with each other.

Low vs high salt intake

1

Steegers et al, 1991: Experimental: low sodium diet containing approximately 20 mmol of sodium per day started in the 14th week of gestation; Control: unrestricted dietary intake

42 healthy nulliparous women with pregnancies < 14th singleton gestation at Nijmegen, The Netherlands

Very small sample size, 14% of the women who were randomized were week excluded from the analysis

Table 2. Characteristics of the seven randomized controlled trials included in the systematic review of balanced protein/energy supplementation during pregnancy with a small-for-gestational age (SGA) outcome

Trial

Type of intervention

Participants and setting

Comments

Blackwell et al, 1973

Experimental: pre-pregnancy and pregnancy supplement containing 40g of milk protein and 800kcal daily plus vitamins and minerals; Control: Supplement containing vitamins and minerals only

294 well-nourished rural Taiwanese women with 'marginal' diets (daily protein intakes, primarily of vegetable origin, not exceeding 40g, only; and energy intakes of approximately 2000 kcal), Taiwan

Interventions 'assigned randomly and blinded' but method not specified; data presented on dietary substitution but based on meal survey discrepancies in first-infant low birth weight rates in 1981 versus 1973 reports

Elwood et al, 1981

Experimental: milk tokens entitling to 1 point of milk a day at half its current price; Control: no intervention

1251 well-nourished pregnant women in two small Welsh towns recruited at time of first reporting of pregnancy, UK

24% of subjects lost to follow-up with evidence of higher losses in control group; higher percentage of smokers in control group

Girija et al, 1984

Experimental: supplement of 50g sesame cake, 40g jaggery and 10g oil, contributing 417 kcal energy and 30 g protein; Control: normal (unsupplemented) diet

20 low socioeconomic status Indian women (monthly family income< 100 US $) in last trimester of pregnancy, India

Large losses to follow-up for breast milk output; no data reported on compliance or dietary substitution; energy and protein intakes appear higher before supplementation even in supplemented group

Kardjati et al, 1988

Experimental: supplement containing 465 kcal energy and 7.1 g protein daily ('high energy'); Control: supplement containing 52 kcal and energy and 6.2 protein daily ('low energy')

747 under-nourished women in 3 villages in Madura, East Java, at 26-28 weeks' gestation, Indonesia

Data on birth weight not analyzed on an intention-to-treat basis; randomization based on household numbers; baseline home diet provided on average 1500 Kcal (±499) 41 g (± 13 g) of protein

Mora et al, 1978, 1979

Experimental:: supplement containing 865 kcal energy and 38.4g protein beginning in 3rd trimester; Control: normal (unsupplemented) diet

456 poor first or second trimester women from Bogota slums for whom at least 50% of previous children had weight-for-height < 85% of Colombian standard

Allocation method not reported; compliance assessed but data not presented; substitution assessed by single 24-hour recall 8 weeks after starting supplement

Rush et al, 1980a,b

Experimental: supplement containing 322 kcal energy, 6g protein, and vitamins/minerals; Control: supplement containing vitamins/ minerals only.

Low-income black women in Harlem, New York City, 'at risk' for low birth weight based on established criteria, USA

Little data presented on the ~ 25% of subjects who failed to comply, dropped out or moved away; stratified randomization based on table of random numbers

Tontisirin et al, 1986

Experimental: one of two supplements containing ~ 350 kcal energy per day and 13g protein (one of the two supplements was similar to women's usual food); Control: normal (unsupplemented) diet

43 healthy Thai women 16-30 years at mean 28 weeks in their first or second pregnancy, Thailand

Number of women originally allocated not stated; no data presented on dietary substitution; several errors in tables

Table 3. Results of systematic reviews of nutritional interventions for the prevention of intrauterine growth retardation

Intervention

Number of trials

Number of participants (E + C)¹

Outcome

Effect²

Forms of care likely to be beneficial

Balanced protein/energy supplementation

7

1423+1353

SGA

0.77 (0.58-1.01)

Forms of care of unknown effectiveness which merit further research

Routine zinc

4

720+698

term-LBW

0.77 (0.54-1.11

Routine folate

5

447+438

term-LBW

0.60 (0.37-0.97)

Routine magnesium

2

678+686

term-LBW

0.59 (0.37-0.93

Forms of care unlikely to be beneficial

Nutritional advice during pregnancy

1

265+250

SGA

1.00 (0.48-2.08)

Isocaloric balanced protein supplementation

1

391+391

SGA

1.61 (1.20-2.14)

High protein supplementation

1

249+256

SGA

1.69 (1.04-2.76)

Vitamin D supplementation

1

59+67

term-LBW

0.50 (0.20-1.26)

Routine iron supplementation

2

1357+1379

term-LBW

0.92 (0.59-1.43)

Prophylactic fish oil

1

113+119

BW < 3rd

0.89 (0.48-1.64)

Fish oil versus placebo

1

266+131

MBW

67g³ (-43.88 to 177.88g)

Fish oil versus olive oil

1

266+136

MBW

126g³ (19.35 to 232.65g)

Routine calcium for prevention of hypertensive disorders

5

774+794

BW < 10th

0.77 (0.51-1.16)

Low vs high salt intake for prevention of hypertensive disorders

1

17+19

BW < 10th

1.78 (0.27-11.52)

¹Experimental and control groups
²Typical odds ratios (OR) and 95% confidence intervals (an OR < 1 indicates that the intervention resulted in favourable outcomes) (Chalmers, 1989)
³Weighted mean difference between intervention and control groups


Figure 1. Effect of balanced protein/energy supplementation in pregnancy on small-for-gestational-age outcome (from Kramer, 1997).

Concerning antianemic supplements, Table 1 summarizes the characteristics of the 5 trials included in the systematic review on routine folate supplementation (Fleming et al, 1968; Baumslag et al, 1970; Giles et al, the 1971; Iyengar, 1971; Rolschau et al, 1979). Although the typical odds ratio shows a statistically significant reduction in the incidence of term-LBW ('Table 3), most of the trials, conducted in the 1960s, defined their populations poorly and did not give details of randomization procedures. All of these trials were performed in populations where iron supplementation was routine.

Routine magnesium supplementation (15 mmol/day) seems to decrease the incidence of term-LBW (Table 3) but this evidence is based solely on two trials, both conducted in developed countries (Kovacs et al, 1988; Spatling and Spatling, 1988). The trials have either weaknesses in randomization procedures (such as allocation of subjects based on the women’s even or odd birthdates) or have high number of exclusions after randomization which make the results inconclusive. There is one other trial conducted in USA by Sibai et al (1989) among 400 young normotensive primigravid women between 13 and 24 weeks' gestation that has not been included in the systematic review. The results of this trial showed no significant differences between the two groups regarding fetal growth retardation.

2.3. Nutritional interventions unlikely to be beneficial

Of the 4 trials included in the systematic review on nutritional advice in pregnancy, data concerning effects on pregnancy outcome are available only from the trial by Kafatos et al (1989). The results of this trial indicate that nutrition counseling during pregnancy improved dietary intake, but did not show an effect in reducing the rate of SGA births (Table 3). The overall methodological quality of the trial is, however, not high. Although random, unbiased allocation is documented, women randomized to receive advice had considerably higher energy intakes prior to randomization than did control women. Similarly, the analysis is based on individual women despite the fact that randomization was done by clinic. There are also discrepancies in sample sizes of related outcomes (e.g., birth weight versus LBW rate), and inconsistent results such as higher head and chest circumferences but no difference in birth weight.

Isocaloric balanced protein supplementation indicates a supplement in which the protein content is "balanced" (i.e., provides < 25% of its total energy content, and replaces an equivalent amount of energy in the diet). The trials performed so far have not compared such a replacement diet with the usual diet, but rather compared the effects of energetically equivalent supplements that differed only in protein content. This systematic review includes three trials, but the meta-analysis principally reflects the data from the large Mardones trial (Mardones-Santander et al, 1988). This trial, however, suffered from large, but apparently unbiased, losses to follow-up and, more importantly, the normal-protein control supplement contained much higher quantities of vitamins, iron, and several other micronutrients than the higher-protein supplement. To date there is, nevertheless, lack of evidence that these micronutrients affect the outcomes examined. Overall, the results suggest an increased risk of SGA births (Table 3); data are insufficient to exclude effects on fetal or neonatal mortality.

The high protein supplementation systematic review includes two trials (Rush et al, 1980b; Iyengar, 1967) but the perinatal outcomes are dominated by the results of the larger Harlem trial (Rush et al, 1980a,b). This is a high quality trial, which unexpectedly suggest the possibility of an increase in the rate of SGA (Table 3). At one-year follow-up, high-protein supplementation was not associated with a detectable difference in weight, length, head circumference, or Bayley mental score. The Harlem trial also reported a non-significant increased risk of neonatal death.

Only one of the 2 trials included in the Vitamin D supplementation systematic review reporting clinical outcomes, the London trial (Brooke et al, 1980), provides data on the effects of supplementation on fetal growth. The London trial targeted pregnant women from the Indian subcontinent living in Britain on the assumption that Vitamin D deficiency is more likely to occur in persons whose diet is relatively low in vitamin D (such as vegetarians) and those who have little exposure to sunlight. The result of the trial showed a decrease in the incidence of neonatal hypocalcemia in these vulnerable population and a non-statistically significant effect on fetal growth, but the sample size of the trial was small (Table 3).

The systematic review on routine iron supplementation in pregnancy includes 2 trials, conducted in well nourished women from developed countries to evaluate the effect of routine versus selective supplementation (Dawson and McGanity, 1987; Hemminki and Rimpel1991). Routine iron supplementation improved maternal iron stores, as reflected by increased serum ferritin and hemoglobin levels, but there were no differences on any substantive clinical outcomes (Table 3). The only possible benefit from routine supplementation, suggested by the larger multicenter trial in Finland (Hemminki and Rimpel1991), was a lower caesarean section rate and postpartum blood transfusion. However, these unexpected findings may be explained by the nonblinding of the trial that may have led to closer observation, possibly bias in measurements, and obstetricians' and midwives' overreactions when faced with low hematocrit and hemoglobin values in the unsupplemented group.

There have been two trials conducted to assess the effects of fish oil supplementation during pregnancy reporting fetal growth outcomes (Olsen et al, 1992; Onwude et al 1995). The larger trial, conducted in Denmark by Olsen et al (1992), is a well designed trial in which the control group received either olive oil, or no oil. Results are presented in Table 3 for the two groups separately. There is some evidence in this trial that the women allocated to olive oil were less likely to increase their fish consumption that women allocated to no oil, and it is also possible that olive oil may have effects that have so far been unreported. The study by Onwude et al is a smaller trial conducted in the UK to determine whether n-3 fatty acid prophylaxis is beneficial in high-risk pregnancies. The data for this trial included in the comprehensive overview (Gglu et al 1997) referred to an earlier report of the trial in an abstract form (Onwude et al, 1992); the results presented in this paper (Table 3) correspond to those reported in the final publication of the trial (Onwude et al, 1995). Based on the limited information provided by these two trials there is currently insufficient evidence for any conclusions to be drawn about the effects of fish oil supplementation during pregnancy in IUGR.

Because of its association with pre-eclampsia and eclampsia, pregnancy-induced hypertensive disorders are important contributors to maternal and fetal morbidity There have been two approaches tested in RCTs to prevent hypertensive disorders during pregnancy that involve nutritional interventions: routine calcium supplementation and low versus high salt intake. The result of the 5 calcium supplementation trials included in this systematic review (Villar et al, 1987; Villar and Repke, 1990b; Belizan et al 1991; Cong et al, 1993; Sanchez-Ramos et al, 1994) did no show any significant effect on preventing term-LBW or SGA (Table 3). There has been only one follow-up into childhood of those babies whose mothers were supplemented with calcium and the results show no long-term effect on physical growth (Belizan et al, 1997). The only trial included in the systematic review of low versus high salt intake in pregnancy is small and, in addition, 14% of the women who were randomized were excluded from the analysis (Steegers et al, 1991). This single trial is to small to provide any trustworthy estimates of the effect of low salt compared to high salt intake during pregnancy on fetal growth (Table 3).